Apply function to each row, with condition-CodePudding

I don't understand why this piece of code does not work. I feel like I'm missing a very simple concept.

#if(!require(tidyverse)) install.packages("tidyverse")

df <- tibble(rule = c("add", "add", "add", "sub"), var1 = c(1, 2, 3, 4), var2 = c(2, 3, 4, 5), result = 0)

simple_as_hello_world <- function(rule, var1, var2) {
  if (rule == "add") {result = var1   var2}
  else {result = var1 - var2}
}

map(.x = df, .f = simple_as_hello_world)

CodePudding user response：

There's a solution with the dplyr package but setting the columns var1 and var2 as numerical, not character. Just remove the quotation marks, or use as.numeric() function.

library(dplyr)

df <- tibble(rule = c("add", "add", "add", "sub"), 
             var1 = c(1, 2, 3, 4), 
             var2 = c(2, 3, 4, 5))

df %>% 
  mutate(result = ifelse(rule == "add", var1   var2, var1 - var2))

# A tibble: 4 x 4
  rule   var1  var2 result
  <chr> <dbl> <dbl>  <dbl>
1 add       1     2      3
2 add       2     3      5
3 add       3     4      7
4 sub       4     5     -1

CodePudding user response：

map as noted works over lists or vectors. Also, your function takes three arguments, so pmap would be best to pass these. You can either make your df a list to map over (works same as passing tibble - list output in both cases of only the result):

library(tidyverse)

df <-
  list(
    rule = c("add", "add", "add", "sub"),
    var1 = c("1", "2", "3", "4"),
    var2 = c("2", "3", "4", "5")
  )

simple_as_hello_world <- function(rule, var1, var2) {
  if (rule == "add") {
    result = as.integer(var1)   as.integer(var2)
  }
  else {
    result = as.integer(var1) - as.integer(var2)
  }
  result
}

pmap(df, .f = simple_as_hello_world)
#> [[1]]
#> [1] 3
#> 
#> [[2]]
#> [1] 5
#> 
#> [[3]]
#> [1] 7
#> 
#> [[4]]
#> [1] -1

Or use within mutate to pass relevant elements to your function:

df <-
  tibble(
    rule = c("add", "add", "add", "sub"),
    var1 = c("1", "2", "3", "4"),
    var2 = c("2", "3", "4", "5")
  )

df |>
  mutate(result = pmap_int(list(rule, var1, var2), simple_as_hello_world))
#> # A tibble: 4 × 4
#>   rule  var1  var2  result
#>   <chr> <chr> <chr>  <int>
#> 1 add   1     2          3
#> 2 add   2     3          5
#> 3 add   3     4          7
#> 4 sub   4     5         -1

CodePudding user response：

pmap(df[,-4], simple_as_hello_world)

map works for a vector, whereas pmap (parallel map) iterates over multiple vectors (in this case the columns of the data frame) and runs the function for x[1], y[1], z[1]; then x[2],y[2],z[2] etc.

also, your "var1" and "var2" are not numbers, they are strings. convert to numeric before trying to add them.